Machine Translation (MT): Is post-editing the silver lining?
The growing use of machine translation (MT) has sparked a debate in recent years about the future of the language services industry. The convergence of language technology, specifically translation memory* (TM), with MT has contributed to a noticeable increase in machine translation efficiency. The availability of MT tools like Google Translate Toolkit has further strengthened the belief that MT will replace the need for human translation in the near future.
Problems with machine translation quality
The major issue with MT is the quality of the translation. Although the accuracy of machine translation has improved greatly, the system’s inability to deal with the intricacies of certain languages is a major disadvantage. Google’s MT has been created using the latest innovation in machine translation – Statistical Machine Translation (SMT). SMT combines algorithms and probability indexes to predict the words being used in translation. Google’s sizeable corpus makes it a prime example of ‘language technology convergence in action’, but despite the advanced technology used, translation output is still of mixed quality.
Google itself has acknowledged that its MT cannot match the quality and accuracy that a professional translator can achieve owing to intricate characteristics found in more complex languages. Many translations produced through MT have to be reviewed and edited manually before use.
Post-editing machine translations
Post-editing is the process of proofreading and editing machine translated content. This is designed to remove linguistic errors created by the automated process, and is managed by a human translator/editor who will prepare the final content for use.
There are benefits to this process as it saves time and costs for the client by combining automation with accuracy. A translator who has expertise in a particular field/subject can be used to edit and verify the content and ensure that the client receive the end product ‘right first time’.
Machine translation has certainly cemented its place in the language services industry. It is being used increasingly across a number of industries where cost savings need to be made. MT will also be incorporated into the translator’s toolkit alongside TM to enhance productivity. If professional translators use Google Translate, they contribute professional created bilingual corpus to the MT engines. This will develop MT capabilities to a greater extent.
MT, as it stands now, cannot be relied upon for producing accurate translations. The use of a post-editing is necessary to obtain accurate translations, and we can’t see much changing for the foreseeable future.